AI tell checker.
See exactly what gives it away.
Paste any text and get the specific giveaways highlighted — the stock phrases, the “delve” vocabulary, the sentences that all land at the same length. Free, unlimited, no signup, and it runs entirely in your browser: your text is never uploaded.
Unlimited & free — runs in your browser, your text is never uploaded
What it checks
The checker runs the same measured rules Shlokah’s style engine uses to keep AI output honest: a vocabulary list of words statistically over-represented in LLM text, a library of stock phrases nobody types twice, and structural analysis — sentence-length burstiness (humans vary, models flatline), em-dash density, and contraction rate. Everything is computed from the text itself, deterministically. No probability guesses, no black box.
That’s also why this is not an AI detector. Detectors guess at authorship and get it wrong constantly — flagging non-native speakers, missing lightly edited output. Showing you the concrete patterns is both more honest and more useful: you can actually fix what you can see.
Frequently asked questions
What is an AI tell?
An AI tell is a pattern that fingerprints machine-generated writing: stock phrases like “it's important to note,” over-represented vocabulary like “delve” and “robust,” suspiciously uniform sentence lengths, heavy em-dash use, and stiff uncontracted phrasing (“do not” instead of “don't”). Individually harmless — in clusters, unmistakable.
Is this an AI detector?
No, and deliberately so. AI detectors output a probability guess and are frequently wrong in both directions. This checker does something narrower and honest: it deterministically finds the specific, known giveaway patterns in your text and shows you exactly where they are, so you can fix them.
Is it really free and unlimited?
Yes. The analysis runs entirely in your browser using the same measured rules as Shlokah's style engine — no server round-trip, no account, no word limits, and your text is never uploaded or stored anywhere.
Why does my writing sound like AI even when I wrote it?
AI patterns are leaking back into human writing — people who read a lot of LLM output start reproducing its habits, and some tells (em-dashes, “moreover”) were always part of formal English. That's why a good check shows you the specific patterns instead of issuing a verdict, and why fixing them means sounding like yourself, not just avoiding a list.
How do I fix the tells it finds?
You can edit by hand — replace stock phrases, vary sentence lengths, use contractions. Or paste the text plus one real email you wrote into Shlokah's free humanizer demo, and get it rewritten in your measured personal style with a score against your own writing.
Finding tells is the easy half
Removing them without losing your voice is the hard half. Shlokah rewrites AI text in your measured personal style — try it on one draft, free.
Humanize it in your voice